Strategic alliances-optimistic fiction or negative fact

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 No.2 MAY 1995

STRATEGIC ALLIANCES:Optimistic Fiction or NegativeFact?Carl-Henric Nilsson"Introduction

Strategic alliances (SA) have bccome increasingly popular as a strategic option inindustry and as a research area for scholars during the last dccadc. Morc strategicalliances have been formed since 1981 than in all previous years (Anderson, 1990),and

"Allianccs have be come an integral part of contemporary strategicthinking" (Sherman, 1992, p77).

The literature reports an increasing frequency of alliance formation (Harrigan,1987; Dowling, 1993) and also prediets inat this trend will continuc, and thai thealliances will bccome globally oricntatcd (Lci and Slocum, 1991). Global strategicpartnerships have

"become an important new strategic option that touches every sector ofthe world e conomy . from sunri se to s unset in dustrie s, frommanufacturing 10 services" (Pcrlmuttcr and Heenan, 1986, p.136).

Ohmae (1989) goes one step further by stating that globalisation makes strategicalliances absolutcly essentialto corporate stratcgy.

Such generaloptimistic opinions are preserned primarily by authors usingcxamplcs of successful alliances as thcir fundamental sourcc of cmpirical cvidcncc(eg Robens, 1992; Spiegel, 1993). This rather glossy picture is howevorcontradicted by others (eg Takac and Singh, 1992), who remind the reader thatproblems do exist in strategic alliances and that thcy can be difficu1t to overcome.Empirical research also points in this dircction. Harrigan (1988) made one of themost extensive survevs of alliance success, exarnining 895 different strategicalliances from 1924 to 1985. She found that in 54.7% of the alliances thecxpcctations of at lcast onc of the alliance partners were not mct. Scientific cascstudies support thcsc conclusions and also present risks involvcd in cc-operation,such as diffusion of care skilIs to alliance partners (1Iamel, 1991).

A doser inspccuon rcvcals thai the picturc convcycd by the aggregatcd literatureon strategic alliances might not be a truc rcflcction of the realnies of the businessworld. As an cxamplc, we can contrast Harrigans findings of an approxirnatcIycqual share of succcssful and unsucccssful allianccx with the findings of the pre~;cf\1analysis, where over four limes as many surcessful as unsucccssful cxamplcs 01allianccx wcre menrioned.

I

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 No.2 MAY 1995

There scems to exist a great divide bctwcen, on the one hand, the optirnistic feelingand rosy picture concerning strategic allianccs conveycd in the literature and, onthe other hand, the empirical Iacts concerning allianccs. The fact that a failedrelationship

"isn' t the kind of thing you like 10 write up in a press release"(Dc Young and Davis, 1990, p36)

is of course one explanation of the lack of empiricaI evidence of failures, but this isnot sufficient to explain the divide and the exccedingly positive auitudcs towardsstrategic alliances.

The mai n purpose of this articlc is lO analyse. in a systematic way, to what externthcrc actually cxists a dividc bctween the optimistic fiction on strategic allianccspresenled in the majority of the literature and the empirically grounded facts ofhigh Iailure rates presenred in other articlcs. A sccondary purpose is to examine ifthe divide can be cxplaincd in terms of author profession, 'scicntificncss' andcmpirical grouriding.

The methodology used is an intcgrativc research review, or meta-analysis, of 121articles, with strategic alliance in the titlc, found in ABI-Inform from 1971 up to1993. The schcmc of the article is as follows. The literature on allianccs and otherintcrmediary forms is rcv icwcd with the pcrspcciivc of co-opcration as analternative, or rather complcmcnt, to cornpctition. Thcn the possiblc divide bctwccnthe optimistic and negative cmpirical literature is outlincd. Following this themethodology is presenled and thercaftcr the resuIts of the rncta-analysis areprescntcd and disenssed. Finally, conclusions are drawn and suggestions for Iuturcresearch preserned.

Competition And Cc-operation

Compctitivc strategy is, by tradition, viewcd from the pcrspcctivc of a companycompcting in a marker against its rivals. Il is bascd on the company' s ability toproteet information and conccal its intentions fromthe compeution (Turner, 1988),thus retaining the company's compctitive edge, Cc-operative arrangements such asstrategic alliances, howover. lift the lcvcl of analysis to a collcctive, or intcr-company, levcl. Bresser (J 988) ar gucs that, under certaln circumstances,companies can usc collcctivc stratcgics in combination with compctitivc stratcgicsto achicvc common goaIs. Howcvcr, the fundamental tension bctwccn cc-operationand collaboratian rcmains and vanes according to his typology. O'Bricn and Tullis(1989) suggest that the boundaries bet wccn compctition and ca-operation areshifting in favour of co-opcrarion. The increasing interest in co-opcrauvcarrangements, such as strategic alliances, can be explaincd by this shift, But whyare the boundarics shirting in the first place?

O' Brien and Tullis suggest a series of trends which will gain Il iomc ntU 111 andIurthcr shift the prcrequisitcs for companics in the 90's: incrc.asing costs of R&.Dand the cornmcrcialisatiou phasc, incrcasing scalc of industrial projccis dcmanding3 ver y large or even global marker for the product, shortcr product lifc eyelex whirh

Journal of ----

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

make time an irnportant factor, political changes and changes in barriers bctweendifferent markets, making it necessary 10 adapt rapidly to a turbulent environment.

Harrigan (1988a) maintains that strategic allianccs reprcsent a significant changc inindustry structure as weil as in the competitive bchaviour of firms. In order not 10be beaten by the cornpeution, and considering the company's environrnent as acomplex, uncertain world filled with dangerous opponents, Ohmae (1989, p143)suggests: "It' s best not to go it alone" .

According 10 Lewis (1992) the best finns are sweeping past their rivals, in SCClOraf ter sector, with strategic alliances. Managers have to ask themselvcs if thcre isnot some truth in the old saying: "I/you can't beat them join them" .

Optimistic Literature On Alliances

There is a grave risk of bcing overrun by co-opcrating competitors, so just as aprecaution, companies should join forces to block off the competition. Yourcompetitors must be regardcd as potential strategic resources.

"Strategic alliances make it possible /O dr aw on excellence fromanywhere in the world" (Mason, 1993, pIO).

Thcrc is no indisputable definition of a strategic alliancc in the literature. Howcvcr,thcrc is an implicit understanding among the authors, of what a strategic a1liance is.Il is implicit since two thirds of the authors do not present thcir definition, ahhoughthere exists an understanding since the discussion in the literature docs not indicateany grave difference of opinion concerning the topic of discussion. In order toindicate what we mean by a strategic alliance, we present a working definition. Astrategic aIliance must ful fil two major criteria: it has to be an alliance, ie morethan one company involvcd, excluding mergers and acquisitions, and the alliancemust be strategic, thai is it must have a substantial impact on each participaLingcornpany's long-term goal, thus excluding, for instance, short-term supplieragrcements.

The case for collaboration is "strongcr dun cvcr" (Harncl, Doz and Prahalad, 1989,pl33). If strategic alliances are so favourablc, what are the induccmcnts? Thrcegroups of morives can be distinguished: efficiency, timeliness and flexibility.'

Efficiency Iactors are always relevant to strategic issucs, in both the production anddistribution dimension. Examplcs of cfficicncy factors are: economies of scalc (cgGross and Ncuman, 1989; Lewis, 1990) for insrancc, via new productiontechnology (Lewis. 1990; Modic, 1988); and economics of scope through, forinstancc, marker expansion (Gross and Newman. 1989). Hermart (1988) also rcfcrsto cfficiency Iactors and bases his explanation on transaction cost theory, which iscspecially relevant in strategic allianccs.

The incrcasing speed of lcchnological change is a major driving force for joiningforccs, as is the ncc d of organisational 1carning (Pucik , 19X8).

Journal of

EUROPEANBUSINESS EDUCATION

ValA NO.2 MAY 1995

"Alliances may be scen as a way of short-circuiting the process of skillsacquisition and thus avoiding the opportunity cost of being a perpetual[ollower" (Hamcl, 1991, (99).

Today,

"the penalty for standing still is far higher than the cost of change"(Bower and Hout, 1988, p l l l ).

Timeliness. manifested primarily as speed to market is demandcd duc to shenerproduct life eyeles (Merrifield, 1992, p77), and this narrows the market windowtime-wise for a product, The total time in the markel can be incrcascd by bcatingthe competition to the marker or by product innovation, as an extension at the endof the normal product life cycle. Wcimcr, Knill, Modic and Poller (1988), argucthat strategic allianccs and simultaneous engincering are manifestations of one andthe same global force, to cope with the increasing complexity of the world.Simultaneous enginccring can also be a means of reaching the marker quickly.

Flcxibility comes in scveral different flavours, a fact to which surprisingly fewauthors pay aucntion. Neverthclcss, different types of flcxibility are regarded as aprim e motive for alliance formation as weil as a success factor. The discussion onflexibi1ity in the literature follows these two main routes. Firstly, flexihility isviewcd as a motive for forming the strategic alliancc, enhancing the relationsnipwith the customers (Hagedoorn and Schakenraad, 1992). This flexibility can beachievcd in response to marker changes and tcchnological changes (Modic, 1988).Secondly, flexibility is viewed as a succcss factor needed lo forge a strategicalliancc, eithcr vicwcd as flexible structurcs on the alliance lcvcl (Parkhc, 1991), orviewcd from the pcrspective of the individual company (Borys and Jcrnison, 1989;Hagedoorn and Schakenraad, 1992). The all iance is a revitalised organisationgaining 'youthful flexibility'. providcd it is unbound by the administrative routinesof the extant organisation (Olleros and Macdonald, 1988, p 159).

According to the majority of the literature strategic alliances are a great idca. Thecurrent trends are towards a globalisanon of products, tcchnology, compctition andallianccs.

"This trend cannot be reversed and we do not wish [O tum back thisdrive towards the [utur c, with all of ils apportunities ... u (Hahn, 1988,pI9).

Strategic allianccs have become a managcrial panacca, it appcars.

A doser inspcction, howcvcr, rcvcals reasons to qucstion the optimism commonlyfound in the literature. The arneles of ten sccrn to be practically , as opposed toscicnufically, oricntatcd. Furthcrmorc, the articlcs appear to be short and Ircqucnilypublished in non-rcferecd journals. The authors sccrn 10 lean heavily on succcsssrorics and exaroples as cmpirical grounrting. Casc studies of succcssful alli.mcc»are also common, whilc survevs are very rare.

Journal of ---

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

Empirical Research On Alliance

Given the entertained suspicion that part of the literature on strategic alliances is atleast a little over-optirnistic, how do strategic allianccs perform on average? Abroad survey sccms to be the most appropriate method to go about answering thaiquestion.

Porter (1987) studicd the diversification program mes of

"33 lar ge prestigious VS companies over the period 1950-1986 and[ound that most of them had divested many more acquisitions than theyhad kept":

A longitudinal study is argued to be a telling way of studying the success andfailure of straicgics. During the period, the studicd companies made 2644 entriesinto new industrics, The entries are divided into acquisiuons, joint ventures andintemal stan-uns. Joint ventures. which are closcly retared to strategic alliances' isthe smallest group, accounting for 8% of the empirical material, adding up lO justover 200 occurrenccs. The vast majority of the empirical material, 70%, isacquisitions. The stud y provides the opportunity of relating one of the predominanttypcs of alliances to other typcs of expansion possibilities. The results indicate a50% divcstiturc rate for the joint venture expansion aucmpts made up to 1980 anddivcstcd bcfore 1987. The corresponding rates wcrc 44% for intemal start-ups and74% for acquisiuons. We must, howcver, hear in mind that Portcr's study is airnedonly at divcrsification stratcgies, joint venture is the only form of alliance in thestudy, and failure is mcasured by divcstiturc rate, which is highly questionablc(parkhe, 1991).

The success rates of strategic alliances are frequcntly discusscd, but the definitionsof SA success and SA perfonnance vary. Porter 's study is aimcd at measuringdivcstiturc, which may not be a valid mcasure of alliance succcss, since alliancescan be intendcd to have a limited duration in time as weil as bcing long lasting.Anderson (1990) discusses how the pcrformance of a joint venture should beasscsscd. Shc found

"startlingly little information on how (and even if) firms monitor andweigh their joint ventures' performance" (Ibid, p20).

The stud y makes the distinction between indic arors of performance anddeterminants of performaneo. Evaluauon of joint ventures is contrastcd withevaluatjon of wholly owncd subsidiarics and the concJusion is that joint venturesdiffcr in many rcspccts and should thercforc be cvaluatcd according to "a morebalanccd, of/en more subjectivc, approach ... " (Ibid, p29), "peimarilv as stand-alone entities, secking to maximise their own performance not the parents'" (Ibid,pn). Wc must norc that Andcrson's concerns are joint ventures. which are stand-alone cntitics, to agreater dcgrce than strategic allianccs generally are.

Harrigari' s (19RRb) study is Iimitcd to "business acuvities where partners may co-oper at c" (Ib id , p53) thus excluding ac quis iuons and intcr nal starr-ups.Furthcrmorc, this dcfuuuon includes not only joint ventures but also other forms ofintcr-firrn co-opcrauon 895 allianccs wcrc studred wor ldwidc, but thc material has

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 NO.2 MAY 1995

a hcavy American prepondcrancc, with 93% of the alliances having at kast oneAmerican partner.

Alliances are defincd as successful in Harrigan's study if thcy are "mutuallyassessed lO be successjul by their sponsors" (Ibid, pS4). Harrigan, as opposed toAnderson, takes the sponsoring finns' view. The author advocates the idea thatpartner asymrneuics are the driving force for the starting and continuing of analliance. The other partner has something that you don't have, but you nccd,

An overall failure rate of 54.7% for alliances was found. Further analysis is carriedout as a function of industry, exploring the influcnce of the sponsoring Iirrns'national ity, relative asset sizc, and also relatedness; horizorual, vertical or non-rclatcd. The results conccming succcss rates are rncagrc. It

"appears thai partner' s traits and sponsor-venture relationship traits donot offer much expl anat or y power in models of venture survival,duration and success" (Ibid, p70).

Howcvcr, when both sponsor finns wcrc unrclatcd to thcir ven ture the failurc ratefar those ventures was 77%. The empirical basis for this conclusion was, howover.only 6.3% of the studicd alliances.

The failure rates indicate that ihcrc are morc problems associated with strategicalliances than rncet the uncritical eye. Still, wc conclude from the literature thatstrategic alliances cannot be dismissed as just another 'buzz-word' in the strategicmanagement literature. Strategic alliances have the potential of providingcompanies with several of the organisationai criteria needcd to cornpctc.Efficiency, timeliness and flcxibility are among the prominent ones. Howevcr,problems in relation to SAs sccrn to exist to agreater extcnt than the alliancc-optimistic literature acknowledges. The implementation phase is one aspect thatappcars to be much casier to discuss theoretically than execute in a real situation.The problems are amplified by the partner sclection process, involving cross-culturaI aspccts on the individual, company and national levels.

Stratcgic allianccs thus surface, not onlyas the of ten over-optimistic picture ofgelden opportunity painred by the literature on strategic alliances, but also as aharsh reality of organisationai dilemmas. This apparent divide in SA literatureseems to be related to the attitude towards strategic alliances as well as the'scicruificncss' of the aniclcs. Two main dimensions thus build up the analysisrnodcl: firstly, the general auitude towards allianccs in a posiuvc-ncutral-ncgativedimension and sccondly, the cmpirical grounding on which the auiiudc is bascd,here dichotomised as facts vs. fiction.

111e analysis modd in Table I presents the six principlc squarcs combining theart iclcs ' attitudes toward strategic allianccs and the empirical grounding. Eachsquarc is also labclled.

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 No.2 MAY 1995

Fact Fiction

Positive positive optimistrealist

neutral 'neutralist'realist

negative pessimistrealist

Neutral

Negative

TABLE 1

THE ANALYSIS MODEL

In order to analyse the nature of the divide in the literature on strategic alliances, amcta-analysis of the literature on strategic allianccs is the orderly methodologicalapproach, and has conscqucntly been chosen. The dimensions in Table l are takenas a starting point for a dceper probc into the charactcrisucs of the literature onstrategic allianccs.

Methodology

A meta-analytical approach allows for inference between empirical and thcoreticalresearch in a time-effective way not feasibly obtainablc by performing theempirical work oneself. An integrative review is lO

"summarise past research by drawing overall conclusions from manyseparate studies thai are believed 10 address related or identicalhypotheses." (Cooper, 1984, plI).

This makes a mcta-analysis the most suitablc mcthod, considering the purpose ofthis study.

The database ABI-Inform was scarchcd for articlcs on strategic allianccs from thestart of 1971 up lO 1993. Fig l illustrates a stcady increase in the accumulatednumbcr of articles in ABI-Inform. The first arneles conceming strategic alliancesappcarcd in 1983 when strategic alliance is found in the abstract (or keywords orLille) of an articlc, In 1985, the first articlc was published with strategic alliance inthe titlc. From thcn on, the number of articles on strategic alJiances has increascdrapidly. At the end of 1993, the accumulatcd numbcr of arneles in the databasc was816138. Of thcsc. 728, or one out of about l HX)O, had strategic alliance in theabstract (or keywords or titlc) and 132 or approximatc\y lp of those had strategicalliance in the tulc. From Fig l we concludc that the proportion of arnelesaddrcssing strategic alliances is still incrcasing sicadil y.

Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

Accumulated relative frequencles

l

M I0.8 'J

_ 0.7 /• 0.6 r:;: 0.5 ./at 0.4 ._;p

0.3 .'"U .~'~1 ~~

0_ N M .••• In 4ID,.... 4D Ol O ••.•. N C").., .." cc t-- 10 Ot O •.... N C")•.... ,....,...,...,...,....,.... •....••.. el Cl)G) 10 lIiIO1010 eCG en 01 001OlOtOOOtCltoOOt~~OtO~CltmOlOlOl~OtOlO- •...•...•...•............•...•...•........... ..-.-- •....•....•...•...•...•... - •...

--- TOlal ABI-Inlorm

----- SA In abstraCl

-.- •• __• SA In ~de

Ve.f

FIGURE l

ACCUMULATED RELATIVE FREQUENCIES FORARTICLES ABI-INFORM

132 articlcs is a managcable sample for the anaIysis. The ratio of aniclcs with SAin the title to articlcs with SA in the abstract (or keywords or uuc) is relativelyconstant, indicating that the sample is not skcwcd with respcct to publication ycar.This sample could also be cxpcctcd LO be concentrared on articlcs with aIlianccs asthe main topic, thus increasing the 'vaIidity dcnsity' of the material. Exclusion ofduplicates, book reviews and conference reviews leaves 121 articles to be analyscd.Thesc articles have been read and classified' in an iterative manner, generating thefinal classification system. The taxa' of each catcgory' as weil as some newcategories were thus developcd during the coding proccdure.

A classification system should fulfil four objcctives: differentiation, generalisation,idcntification and information retrieval. In order to achieve these objcctives allcatcgorical levels must be: mutually exclusive, internally homogeneous,collcctivc\y cxhaustivc, stablc and bascd on relevant names (Chrisrnan, Hofer andBoulton, 1988, p415). The classification system used in this stud y was constructcdwith this in mind. AJI taxa are munrally exclusive within cach catcgory. Howevor.some of the categorics wcre allowcd to overlap each other in an offort to capturethe fundamental charactcrisucs of cach cntity from scvcral different pcrspcctives.

The validity and reliability of the coding is of fundamental importance for thercsults of the analysis. To ensure proper coding the foIlowing prccautions wcrcapplied. All artic1es were codcd by the same person (the author) and in randomorder. All coding was controllcd at lcast once af ter the final changes to the codinginstrument. This involvcd substaruial rc-coding of the carlicst codcd material duc toelianges in the instrument. All variables were chcckcd for corrclation with thecoding order. No significaru corrclauon was found cxccpt for the coding time for

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 NO.2 MAY 1995

each article. However, this is not rclated 10 the articlc content and was not uscd inthe analysis. It is merely an indication of the effcct of the lcarning curve.

In order 10 control the coding procedure the inter-rater rcliability was calculated. Arandom sample of two articlcs was sclccted and codcd by an extra coder. Thecoding schemes of the regular and extra coder wcre then compared and theincidence of identical coding of the 34 variables of cach article was calculated. Theinter-rater reliability was 82%, which is regarded as a satisfactory lcvcl for theinstrument employed.

The coding instrument is dividcd into four groups of catcgories (or variables, theterms are interchangeablc). In the first group, 'publication' is codcd, for instancc,publication year, author gender, author profession, number of pages, number offigurcs, etc. The seeond group is 'research', where referee proccdure, empiricalgrouriding. methodological discussion, number of rcfcrcnccs, ctc., is coded. Thethird group deal s with 'contcru' of the articlcs, such as: definition of SA, jointventure discusscd, industry and alliance typc. The founh group concerns the'auitudc' toward strategic alliances. This is the group which involvc the mostsubjective appraisals. The aim was, thcrcforc , to build up a hicrarchicalc1assification of thcsc entities in such as way thai the number of taxa at the lowestlcvel of the hierarchy (positive, neutral, negative) was rninirniscd and the richnessof the rncasurc was achieved by aggregating thcsc taxa to a higher lcvcl in thehierarchy.

Attitude Index

Articles can convey messages in more than ene way. The most obvious way isdircct\y and openly via the outspöken statements of the text. But messagos are alsocommunicated indircctly, consciously or unconsciously, via the characteristics ofthe text in which the topic is ernbeddcd. For instancc, if an articlc on strategicalliances opcnly states that there are both advantages and disadvantagcs of strategicallianccs, without giving reference to any empirical material, the text appcarsalliance neutral to the reader. provided that the advantagcs and disadvantagesmentioned are equivalent in frequency and importance. Howcvcr, if in addition tothe abovc, 'an articleflashes out numerous examples of successful allianccsthroughout the text, the reader is indirectly given the impression that strategicalliances are favourablc. This also holds truc for the tcxtual context cncompassingwords synonymous or closcly rclatcd to strategic alliances, such as cc-operation,joint ventures. allianccs and, of coursc, strategic allianccs.

In order to capture the esscnce of an articlc's auitudc towards strategic allianccs, anattitudc measure was developed arter the coding of the aruclcs. Il is aimed atembracing boih the dircct and the indircct messagos of an aniclc. Thrce variablesare involvcd. Firstly, the articlc's 'asscsscd auitudc' to strategic alliances is codcdon a scalc from (l) very negative lO (5) very positive, with O) indicating a neutralauitudc. This measure is aimed al capturing primarily the dircct messago in thearuclc. Sccondly, cvcry time an "cxamplc ' of an alliance is mcnuoncd in the text.the cxamplc is asscxscd as convcying a messago concerning the strategic allianrc:positive, neutral or negative. Thirdly, the latter (lHKCÖUIC is upplied cvcry lime the

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 NO.2 MAY 1995

'term' 'strategic alliance' is mentioned. From these three factors an aggregatedmeasure of auitudc toward strategic alliances, the 'attitudc index', is calculated.

The auitude index was constructed in order \O allow the three components to bewcighted \O form a single measure. Conccptually, it is advantageous if the scale isconstructcd with a range from (-I.(X)) very negative, via (O) completcly neutral \O

(+ 1.00) very positive. The assesscd attitude toward SAs is converted into the'asscssrnent componcnr' having a -1.00 to +1.00 scale by calculating (x-3){l, wherex is the assessed auitude on a fivc-point Likert scale. The 'SA examplc componcnt'is calculated by adding the number of positive, negative and total number ofexamplcs of SA mentioned in the text of the article and computing:

SA examplc cornponcntpositive SA examples - negative SA examplcs

total number of SA examples

The 'SA term component' is calculated analogously \O the example component. Allthrce components thus have a maximum of +1.00, a minimum of -1.00, and O as theneutral point. The thrce cornponents are then weightcd togother to form the'auitudc index' dcfined as:

assessment + SA cxample + SA termAttitude index

3

Equal weighting factors have been used for the thrce cornponcnts, but other weightsare possible and have been tricd. The results of the analysis, howcver, do notehange significantly, hencc equal weighting factors were used in the analysis.'

The statistical tests appl icd in the analysis are rather erude, due to thecharacteristics of the underlying data. The auitude index, for instance, docs notappear to comply with a normal distribution, and too many missing values (egconeeming author profession) inhibit Anova analysis involving this variable. Forvisual display the auitude index is dividcd into thrcc taxa and analysed withcontingency tables. Duc to the skcwncss of the data, cautious interpretation isrcquircd.'

Rcsults

Correlation between au thor profession, 'scicnuficncss", empirical grounding andauitudc index is proposcd. Thcsc rclationships are furthcr tesred via tradiiionalstatisucal analysis. Thirty-Iour variables wcre codcd for cach of the 121 arneles inthe study. All aniclcs togother with abbrcviaicd bibliographical rclcrcnccs are listcdin Appendix A. Furthcrmorc, somc of the most interesting variables of the analystsare also lisred for cach aniclc in the Appendix.

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 No.2 MAY 1995

Thcre is a striking difference betwecn scientific and practical articles. Thescientific articles are of ten written by academics (78%), making refercnces toprevious literature in the field (82%) and discussing the assumptions and themethodology on which the results are based (61 %). Thcse articles are also ratherlong (73% ~ 7 pages) and orten published in refcrecd journals (80%).

In contrast. the practical literature seJdom discusses the assurnptions ormethodology used 10 derive conclusions (99%). Furthermore, the articles are ortenshort (91% ~ 6 pages), without any refercnccs (96%) and are published in non-refereed journals (84%). They are different in most rcspects: by whom they arewritten, and how they are writtcn, and presumabJy they are also wriucn with adifferent purpose in mind and aimcd at a different target rcadcrship,

Industry Consultant Mix Acadcmic (Missing) Totals

Scicntific 6 3 2 38 2 51

Practical 18 Il 6 34 70

Totals 24 14 3 44 36 121

TA8LE 2

FREQUENCIES FOR 'SCIENTIFICNESS' AND AUTHOR PROFESSION

(X" = 33j p<O.OOOlj contingency coefficient = 0.53)

Academics are responsible for the majority of the scientific contributions, whileauthors from industry and consultants are rcsponsiblc for thepractically orientatedarticles. This does not come as a big surprisc. What is more surprising, though, isthe lack of co-authorship bctwccn acadcrnics and non-academics. Such co-operation could have been onc way of bridging the gap bciwccn scientific andpracucal aniclcs.

'Sc icnuficncss' and crnpirical grounding for the articles are two independentvariables which, according to the purpose of this study, are cxpcctcd 10 be rclatcdto the aniclcs auitudc towards strategic allianccs, But how are ihcsc variablesrclaicd LO each other? A contingency table can shcd sornc light on this issuc.

Table 3 rcvcals that almost all survcys are scicnufic publications. Casc studies andnon-cmpirical studies can be both scicruific and practical. A majority of the studiesthaI are bascd solcly on cxamplcs are practical. The table also shows Ulat survcys

Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 No.2 MAY 1995

Examples Non-ernpirical TotalsCaseSurvcy

10 14Scicntific 16 5111

Practical 15 37 17 70

Totals 25 5112 33 121

TABLE 3

FREQUENCIES FOR 'SCIENTIFICNESS' AND EMPIRICALGROUNDING

(X' = 17; p<O.OOl; contingcncy coefficient = 0.35)

are most scarce while examplcs are the most common empirical grounding. In all,practical articles are more common than scientific articles.

Rccall that we expectcd a correlation betwccn an artic\e 's attitude to SA and'scicnuficncss' as weil as the empirical grounding of the article. The auitudc ismcasured by the auitudc index. Thus what we nccd is a third dimension in Table 3containing each artic\e 's rating on the auitudc index. An alternative to a 3Ddiagram is preserned in Fig 2. This cornplcx figure contains much of the esscnce ofthe findings of the study.

Three variables are involvcd in Fig 2: the 'auitude index', which is the dependentvariable, and two independent variables: 'scicntificness' and 'empirical grounding';survcy, casc studies, examples or non-empirical. The two independent variables,with two tirnes four taxa, create eight different groups of articles, the same eightgroups as in Table 3. In Fig 2, the articles within each of the eight groups arearranged in ascending attitude index order: starting with the scientific surveysranging from the survey with the lowest auitude index, which is (O) to the surveywith the highest auitudc index which is (0.69). In the figure, filled symbolsrepresear scientific and opcn symbols practical articlcs. The single opcn circle thusindicates the only pracucally orientated survey with a attitudc index of (0.48).Thereaftcr, case studies, cxamplcs and non-empirical articles are arranged in thesame manner.

Fig 2 makes ii totally clear thai the majority of the literature on strategic allianccs ispositive towards alliances. &4% of the arneles have a positive auiiudc index while13% have a negative anitudc index (3% have zero). If we compare the filled,scicntific symbols with the opcn practical oncs, all articlcs are skewcd towards thepositive end of the sea le , hut the scicntific articles are less skewed.

Journal of ----

EUROPEANBUSINESS EDUCATION

VoI.4 NO.2 MAY 1995

1.00

75

K.50

'"u .25c::

j o~~ .•• "---.;.."..,_..-----;-:A~---.••...c...-------T-I.--:#~--•...---t

:i.50 : i,.·.75

;

·'00 ••• i• Sci, o ~ .• Sci, c Practical ~ $dentific A Practical ~ Scientitic O Practical

SUAVEY , CASE STUDIES EXAMPLES NON·EMPIRICAL

FIGURE 2

THE 121 ARTICLES IN ORDER OF ASCENDING ATTITUDE INDEXACCORDING TO EMPIRICAL GROUNDING AND 'SCIENTIFICNESS'

111e titlc of this study proposes 'opurnisuc fiction' 10 be contrastcd with 'negativeIacis'. From Fig 2 wc conclude thai, whilc opumisuc fiction CXiSLS(inat is, SA-positive aruclcs without empirical backing), the negative facts concerning SA arevery rare. Thus, the situation is rnore cornplcx than the lille suggests. The analysismodcl presenred in Table 1 takes this into consideration. dividing the auuudc indexinta three taxa: positive, neutral and negative. The folIowing presentation reliesprimarily on the data presentcd in Fig 2.

For reasons of visual display the material is divided inta ordinal taxa and presernedin contingency tables. The position of the lines di viding the attitudc index intathrce groups can arguably be placed at different locations. The material is, wirhinrcasonable limits, not espccially sensitive LO where the dividing lincs are placed. Asimple and usablc principlc is lo split the scale into Lhrce spans of cqual sizc. Thismeihod is used, defining neutral bctwccn -0.33 and +0.33; negative bclow ·0.33and positive abovc +0.33.

To make use of all the information given by the coruinuous auirudc index, theSpeannan rank corrclation (Spcarman ' s rho) is presenled in the text to support thecontingency data. The Speannan corrclation was ehosen as an alternative to theusual corrclation cocfficicni. Il rcquircs only ordinal data and since il is bascd onthe rank of the data, not the data uselt. it is also rcsistant lO outlicrs. All Spearrrianr.mk cocfficicnts throughout the study are corrccicd for lies and reporred with theconexponding p-valne.

Two major aspeers of Fig 2 can be represemed by Tables 4 and S. Talile 4illustrates thai the dilfcrcncc bctwccn scicnufic and practical articlcs is cOllc);l\cd

to the arriclcs auitudc towarcls str.ucg ic al liancc (Spcarrnans rho "' fl :7, fl <

Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

Scicntific TotalsPractical

24 55

26 10

1 5

Positive 79

Neutral 36

Negative 6

Totals 51 70 121

TABLE4

FREQUENCIES FOR ATTITUDES TO SAs AND 'SCIENTIFICNESS'

(X' = 19; p<O.OOOI; contingency coefficient = 0.37)

OJX)4). The scientific articIes are dividcd approximatcly cqually betwccn positiveand neutral, whilc 79% of the practical articles are rated positive.

A corrclation bctwccn auitude toward SAs and the articlc's empirical grounding isex pee led. From Table 3 and Fig 2 it is elear that an articIe can be based on asurvcy, case studies, examples or can be non-empirical. But what characterisesthese taxa and how different are they?

One third of the surveys are bascd on 100 to 600 alliances, and another third arebascd on rnorc than 600 studicd alliances. Casc studies, on the other hand, arebascd on fcwer alliances, but eonsritute contcxtually deepcr studies. Two thirds ofthe ease studies are bascd on a single casc, while just one out of twenty-fivc casestudies is based on morc than ten cascs. Authors can also draw thcir conclusionsbascd on examples. Howcvcr, it is much more diffieult for the reader to tell howdcep the author 's knowledge is conceming a specilie alliance that is merely uscd toexemplifya point. Equally, he might be either quoting hcarsay or deseribingpersonal experience. In general, howcver, the author's knowlcdge of an allianccmentioned as an cxample is rcgardcd to be substantially lowcr than an author 'sknowledgc of an alliance in a case study. The number of exaraples per articlc isfairly evcnly distributcd bciwccn onc and twcnty-fivc cxamplcs, but morc authorsusc fcwcr eX3lT1pks_One outlicr uscs as manyas fort y-nine different cxarnplcs ofallianccs. -111efourth and last typc of grounding is non-cmpirical.

Table 5 illustrates auitudc towards SA as a funetion of empirical grounding. Tenout of twcl ve survcys are ratcd to have a neutral auiiudc. Thcse studies basc theirconclusions 011 the nornoihctically' most finn ground. The conclusions from asurvey are, on avcragc, bascd on 393 alliances. The cquivalcnt nurnbcr of alliancesis 4 in casc studies, l I in ex amples and of course zero for the non-crnpiricalariiclcs. The thrcc latter groups have thcir central point in the positive squarc withapproximatcly 7()'J(, of the arneles ratcd as positive.

Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

TOlals

Positive

Neutral

Negative

Survey C ase Examplcs Non-cmpirical Totals

2 18 37 22

10 7 Il 8

O O 3 3

79

36

6

12 25 51 33 121

TABLE S

FREQUENCIES OF ATTITUDES TO SAs AND EMPIRICALGROUNDlNG

(X' = 21; p<O.002; contingency coefficicnt = 0.38)

Spcarman's rho = 0.09, p = 0.33 in Table 5, indicating that no linear rclationshipcxists, while X' indicates same sort of relationship (X' tests are inscnsiuve to theeffccts of order). The dilemma is solved by the Anova plot in Fig 3, showing theintetaction line plot for the attitude index with the cffccts of 'scicntificncss' andempirical grounding. The error bars indicatc the 95% confideneo intcrval.

.80

".70.,

"O .60.s'" .50"O

-~.40;(

ö .30a>O> 20~.,> 10«

0.00

10

-e- Practical___ Scientific

f-L-__ ~ ~ ~ ~ _

Survoy Caso studies Non-cmpmcalExarriples

L.... _

FIGURE 3

INTERACTlON LINE SIIOWING THE VARIATION IN TIIE I\1EANVALUE OF THE ATrITUDE INDEX SPLIT BY 'SCIENTIFICNESS' AND

EMPIRICAL GROlJNOING

Journal of ---

EUROPEANBUSINESS EDUCATION

Vo1.4No.2 MAY 1995

However, Fig 3 should be interpreted cautiously. The data contain just one singlepractical survey. It is interesting to note that the scientific articles' auitude indexincreases with decreasing strcngth of empirical grouriding. which is fully logical,while the practical articles show the opposite pauern (with the exception of the onepractical survey). This is the cxplanation of the low correlation according LO

Spearman's rho: the practical and scientific articles neutralise each other.According LO the Anova analysis, significant differences in the means at the 5%level, according to Fisher 's protectcd least significant difference, are found for:survey vs. case studies, mean diff = 0.29 and for survey vs. examplcs, mean diff =2.5.

Table S, which indicated a low correlation, can thus be further split by"scientificncss ', generating Tables 6a and 6b. These tables rcveal the dividebctween the optimistic fiction regarding strategic alliances preserned in themajority of the literature and the cmpirically grounded facts of high failure rates ofstrategic allianccs, preserned in other articlcs. Spcarman's rho is: for table 6a 0.28and p<O.OS;for table 6b -0.22, and p<0.07 thus indicating a higher significancethan in Table 5.

Case Non-ernpirical TotalsExarnplesSurvey

Negative

l 5 7 11

la 5 7 4

O o O l

26

Positive 24Neutral

Totals la 1614Il 51

lABLE6A

FREQUENCIES BASED ON SCIENTIFIC ARTICLES' GROUNDING ANDATTITUDE TO SAs

(Xl = 13; p<O.05j contingency coefficient = 0.45)

Casc Examplcs Non-cmpirical TotalsSurvcy

l n 30 Il

o 2 4 4

o o 3 2

Positive 55

Neutral 10

Negative 5

TOlals J715 37 70

TABLE6B

FREQUENClES BASEn ON PRACTICAL ARTICLES' GROtINDING ANDATTITlJDE TO SAs

(X' i., 4; p<O.07; contingenq coefficient = 0.23)

Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

The scientific articles are not as positively skewed as the practical literature. Onepossible explanauon for this is that the practically orientated literature usesexarnples and cases as the primary source of empirical data and, as was mentionedin the introduction, there is probably an access problem hindering the stud Y ofunsuccessful strategic alliances. Interestingly enough, the writers of the scicntificpapcrs seem to manage the access problem better since they are distributed equallybctween positive and neutral for case studies and examples. Nearly all survcys, themost probable method of obtaining an unbiased result in the attitude dimension, arescientific articles and ten out of eleven of these articles are rated neutral on theattitude index. Of these ten articles, seven were writtcn by acadcmics, one by amixture of acadcmics and non-acadernics, one by a consultant and one was wriuenby an author from industry.

The major boundary between empirical grouriding. labelled as fact and fiction inthe analysis model, could be argucd to be anywhere bctween survey and non-empirieal in the empirieal grounding dimension. Il could, for instance. be betwccnempirical and non-empirical studies, embraeing survcys, case studies and examplcsin the empirical group. It could also be argued lo be bctwccn survcys and ca sestudies on the onc hand, bcing firmly grounded, and cxamplcs and non-cmpiricalon the other hand, as anccdotal. Finally, the dividc can be argucd to be bctwccnsurveys and non-surveys, From Table 7 il is elear that the material suggests adi vide bctwecn survcys and non-survcy studies. A closcr examination from anomolhetic point of view also reveals lhat all surveys are bascd on 50 alliances ormore, while all non-surveys are bascd on less than 50 alliances.

Positive 79

Eramples ornon-empirical

Surveyor case

Survey, cascorcxamplcs

Non-Survey Survcy Non-empirical Totals

2 77 20 59 57 22

10 26 17 19 28 8

O 6 O 6 3 3

Neutral 36

Negative 6

Totals

X' = 18p < 0.0001

37 84 88 33

X' = 8 X' =2p < 0.02 p < 0.4

= 0.08 = om< 0.39 <0.92

10912 121

Spearrnan's rhocorrectcd for tics:Ticd p-value:

= 0.25<0_001

TABLE 7a, 7b, 7c

FREQUENCIES FOR ATTITUDE TO SA FOR TIIREE DIFFERENTDlCHOTOMISATIONS OF EMPIRICAL GROUNOING

Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 No.2 MAY 1995

Thus, retuming to the analysis mode! in Table l, survey represents facts in thisstudy and fiction is represented by articles based on case studies, examples andnon-empirical research. This is illustrated in Table 7a which prov ides the clearestdivide scientifically and empirically. Case studies and surveys do not bclong to thesame group, as proposcd in Table 7b. Case studies are not providing the properscientific grounding for conclusions regarding the general success of strategicalliances. From a scientific point of view, this is natural since case studies are mostapt for empirical induction and theory generative purposes, in contrast to surveyswhich are deductive and more apt for testing general patterns and verifyinghypotheses. Examples as empirical grounding see m questionable based on theresults of this study. Non-empirical studies also secrn questionable. The time trendof the attitude index is weak in the material. But while the time trend of the auitudeindex is negative for survcys, the trend is positive and stronger for the non-empirical studies.

The fictional literature. here represenred by the non-survey articles, can act as aninsugater for strategic alliances in providing decision makers with idcas and insightthat strategic alliances can be favourable or even very favourable. The factualliterature. represenred by the surveys, on the other hand, can provide objcctiveinformation conceming alliances. The task of asscssing the valid ity of the materialpresented in the literature thus differs depcnding on the empirical grounding of thearticle.

Survcy Non-survcy

Positive 50% 32%

90% 58%

- 30%

Neutral

Negative

TABLE8

THE PERCENTAGE OF REFEREED ARTlCLES FOR EACH OF THE SIXSQUARES OF THE ANALYSIS MODEL (-BASED ON ONLY TWO

POSITIVE SURVEYS)

In the Iactual literature (surveys) most artic1es have been screened beforepublication. thus ensuring validation by the externai referee proccdure. Thepositive survey square contains only two artic1es, thus making il difficult to drawany valid conclusions conceming this square. All but one of the neutral surveyswere rcferccd. In the non-survey studies, especially the positive and negative oncs,the task of judging the rclevance and validity of the articles is lcft primarily to thercader. A larger proportion of referecd articles is round in the neutral squarc (58%)

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 No.2 MAY 1995

than in the positive and negative non-survey squares (32% and 30%). One possibleexplanation for this is that il is difficult to convince the referees of all the rosypictures of alliances, thus restrioting these articles to non-rcfcreed journals.

A further hypothesis is that while the factual articles are usually intendcd to conveya correct picture of strategic alliances and their pros and eons, the practicallyorientated literature is not bound by this objective. Instead, fictional literature isintended to convey a correct picture of the aspects of reality tnat the author choosesto present, which may be the author's expericnce of certain strategic alliances. Ifthis is true, it is an additional indication to the readers of the literature on strategicalliance that they must thernselves assess the validity of any general conclusionsconceming aIliances presented in these studies.

Dividing the analysis in Table 7 further into scientific and practical artic1es doesnot enhance the results. The major divide is found in Table 7a, between theoptimists (77 articles) and the neutral realists (10 artic1es). The rnatrix also showsthat 'neutralists' are not unusual, relatively few pessimists and positive realistsexist, while no negative realists are found. (Spcarman's rho = 0.25; p<O.Ol).

Fact = survey Fiction = non-survey

Positivepositive 2 optimist 77realist

neutral 10 'neutralist' 26realist

negative O pessimist 6realist

Neutral

Negative

TABLE9

THE RESULTS OF THE ANALYSIS MODEL. THE NUMRERSINOICATE THE NUMBER OF ARTICLES FOUND IN EACH SQUARE

Discussion

The substantive area of this study is the literature on strategic alliances not theallianccs thcmselvcs. Being without empirical grounding il would be hazardous todraw conclusions from this study concerning strategic allianccs themselves.Conc1usions conceming the literature on allianccs, however, are firmly groundcd.

The empirical basis of a study can be expectcd to be biased depcnding on thernethodology used. This suspicion is strongthened by the discovered skewness of

I Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 NO.2 MAY 1995

the material. Surveys are argucd to have least potential for bias since the surveyrespondents often are protectcd by anonymity. Surveys are further of ten answeredin a writtcn form. It is probably easier 10 admit a failure in a written anonymousform than to do so in, for instance, an interview situation. Most surveys alsoindicate a failure rate of approximately 50%, indicating an equal number ofposiuvc/favourablc alliances and ncgative/unfavourablc ones.

In case studies, the distribution of positive and negative alliances may be biased fordifferent reasons. Firstly, there may be an 'access bias'. The potential respondentsare asked to participate in a case study. Those respondents that can showaposiuve/tavourable alliance are probably more wilIing, as are their superiors, toparticipate in the case study. A bias towards successful allianccs could thus beexpectcd. Sccondly, there may be a 'response bias'. Those respondents that arewilling 10 lake part in the study are probably som chow rcsponsible for thcir allianceand want to dcscribe their alliance, eonsclously or subconsciously, in a favourablelight. Thirdly, there is a potential for 'selcction bias' by the au thor. In the cascstudies analysed in this study six success cases are studied for each failure.Assuming the failure rate of approximatcly 50% found in broad surveys, thisindicates a clcar bias in the case studies. The practical case studies are responsiblefor the lion's share of the potential bias, since they do not study any unfavourablealliances at all.

A similar bias is also possible for studies using exarnplcs, A positive 'sclectionbias' can be expcctcd if the examplcs are sclcctcd to confirm the author'shypotheses. In addition to this, 'access bias' is possiblc. If the sources are rcluctantto recount incidents of failure, the available stock of alIiances from which tochoose exarnplcs is probably already biased towards positive alliances. The studiesbased on examples presented more than four exarnplcs of successful strategicalliances for each failure, indicating a probable bias in articles based on examplcs,especially practical articles.

Bcing biased without any empirical backing would be a 'prejudiee bias'. The non-empirical aruclcs, however, usc the literature on strategic alliances for references.This literature may be biased, as discussed above. The non-empirical artic1es maythus also be biased, and if they are, ilwould be expccted lo be towards successfulalliances. Admiuedly, the bias of non-empirical articles is somewhat spcculativeand it is not possible to confirm it with the data in this study in the same way as forcase studies and exaraples. However, an indication of possible biases as a functionof 'scicntificncss' and empirical grounding is given by the skcwness of the avcrageauitudc index shown in Fig 3.

Conclusions

This study empirically confirms three major issues. Firstly, the literature onstrategic alliances is, on average, over-optirnistic. Sccondly, thcre exist clearrclationships bctwccn author profession, 'scienuficncss', empirical grounding andauitudc toward alliances. Thirdly, the natural di vide betwecn different types ofempirical grounding is bctwccn survey and non-survey studies. This divide may befurther cxplaincd as a function of the biases intrinsic to each method used. Theskcwness of the material is taken as a probable indicator of these biases. Survcys

Journal of

are least skewed thus indicating less bias. The case studies and exarnples can beskewed duc to access bias, response bias and selcction bias. The non-empiricalstudies can be skewed due to prcjudice bias. Probable biases were found for allmethodological approaches and were higher for practical studies ihan scientificones, with the exception of non-empirical studies where the scientific and practicalstudies wcre approximately equally skcwed.

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

The stud y indicates that two major groups or articles exist: the optimists,representcd by non-survey-based articles ratcd positive on the attitude index, andthe neutral realists, representcd by surveys ratcd neutral on the attitude index.Three minor groups also ex ist: 'neutralists' are not unusual, relatively fewpessimists and positive realists exist, while no negative realists are found.

From the literature. il appears that the potential of strategic alliances is overratcdwhile significant realisation difficulties are underratcd. Thcrefore, it should be ofinterest for practitioners, as weil as scholars, to el an fy the potential of strategicalliances togother with the realisation difficulties in order to bridge this dividc. Co-authorship bctwccn academics and non-academics is one way of reachingpractically applicable implications based on the cxtant literature that is empiricallygrounded in more than mere exarnplcs of successful alliances. Surprisingly fewsuch studies have been found.

Studies of other than successful alliances could prov ide important insight into thedeterminants of formation of succcssful alliances. For instance, casc studies offailed alliances are e1early underreprcsentcd in the literature. Funhermore,remarkably few articlcs discuss the possibility of not taking part in an in analliance. In order to understand the advantages of taking part it would beadvantagcous to also vie w alliances from the perspective of a company lhalconsciously ehooses not to lake part. Thcrcfore, studies of companies that haveehosen not to lake part could also offer a means of gaining important insight intothe pros and, espccially, the eons of all iancc formation.

The rcsults of this study could be improvcd by reducing the lacking informationconcerning. Iorinstance, author profession in order to validate these results as weilas refine them via Anova analysis, or, in order to lake into account the intcractioneffects, analyse the material with LISREL. Furthermore, similar analyses couldalso be performed on other relevant issucs. such as joint ventures, merger andacquisitions to name some potential topics.

In this article a sobering picture of strategic alliances is paintcd. Taking part instrategic alliances can be an advantagcous strategic move for some cornpanics, butin other cascs it may still be belter to heat thcrn, than to join thcrn.

NOTES

l. There are of course other types of alliance monves than thcsc thrccgroups. For instance, Bresser (1988) argues from a theorctical point ofview that the major motivation for collaborative strategles is the nccd toreducc uncertainty in the cornpany's environment.

Journal of

EUROPEANBUSINESS EDUCATION

VoI.4 NO.2 MAY 1995

2. For a discussion on the differences between joint ventures and strategicalliances, see Baranson, 1990.

3. "Classification is the development of a system or scheme in order forresearchers to arrange entities into taxa based on their similarities,differences, and relationships to one another as determined by orinferred from thcir most fundamental eharacteristics" (Chrisman, Hoferand Boulton, 1988, p415).

4. Taxa are "sets of entities sufficiently sim ilar to each other andsufficiently different from the entities in other such scts that they areseparately delimited and named" (Ibid.).

5. "A catcgory is a rank or a level in a hierarchical classification which iscomposed of taxa to which a given rank is assigned" (Ibid.).

6. Of the 121 aruclcs, 32 lack one of the three components constituting theauitude index, making an analysis of missing valucs ncccssary. For thearticlcs with missing values the asscsscd attitude to SA was not missingin any ease. The missing valucs were in either the examplcs or the terms,either due to the fact that the authors did not mention any exarnplcs (28articles), or did not rncntion the term strategic allianccs (4 articles). Themissing valucs wcrc significantly related to the articlc grounding (xl=S4)making re-coding of the missing values nccessary. An unpaired t-testrevealed an insignificant difference (p<O.S) in the means of theparameters bctwcen original and missing. Thcrcfore, the missing valucswere prcdieted bascd on the two values available.

7. The dependent variable in many of the analyses, the auitude index, isskewed towards the positive end of the scale. This creates somestatistical problems. Firstly, the material is not normally distributcd, thusreducing the applieable statistical arsenal. Secondly, although r testsallow different types of distribution, the classifieation table has to fulfilcertain minimum criteria, such as expected values of at least 5 in at least80% of the squares, and no square with an expected value of less thanonc, (For further discussion see eg Siegal, 1956). Several of the rnatricesin this analysis are cIose to, but do not fuIIy comply with thcse eriteria.For these analyses, individual cell xl values have been caIculated inorder to analyse the impaet on the total r. In all eascs, the eontributionfrom the questionable cells to the total r value was low, although notinsignificant. In the contingeney tablcs the total r is still given but, inaeeordance with the discussion above, must be interpreted cautiously. Insome eases, the contingeney eocfficient is reported. This measurc is theequivalent of a correlation coefficient for contingeney tables. It onlyrequires nominal variables and hence does not eonsider the order ofvalues within rows and columns.

8. Nomothetie is here defined as the methodological approach whichemphasises "the importance of basing research upon systematic protocoland tcchniquc" (BurrelI and Morgan, 1979, p6), using quantitative

Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

techniqucs for the analysis of data collected, for instance, by surveys.The oppositc approach is the idcographic approach, based on the viewthat one can only understand the social world by "getting inside'situations and "letting one's subject unfold its nature and charactcristicsduring the process under investigation" (Ibid). For a discussion on thismethodological debate, emanating from the literature on socialphilosophy, see BurrelI and Morgan (1979).

9. I would like to thank Ingvar Persson, Lars Bengtsson, Sven Coli in,Agneta Planandcr and Kristina Henriksson for comments on earIierdrafts of this paper. I would especiaIly like 10 express my grautude 10Rikard Larsson for several commerus and suggestions 'beyond the callof duty' on this papcr. This study has in part been fundcd by the Bank ofSweden Tercentenary Foundation.

REFERENCES

ANDERSON, Erin (1990) 'Two Firrns, One Frontier: On Assessing Joint VenturePcrformancc', Sloan Management Review, Winter, ppI9-30.

BARANSON, Jack (1990) 'Transnational Stratcgic Alliances: Why, What, Whereand How', Multinational Business, No.2, pp54-61.

BORYS, Bryan & JEMISON, David, B. (1989) 'Hybrid Arrangements as StratcgicAlliances: Theorctical Issucs in Organisational Cornbinations', Academy ofManagement Review, Vol 14, No.2, pp234-249.

BOWER, Joseph L. & HOUT, Thomas, M. (1988) 'Fast-CycIe Capability forCompetitive Power', Harvard Business Review, Nov/Dec, ppl10-118.

BRESSER, Rudi K.F. (1988) 'Matching Collectivc and Compctitivc Strategics',Strategic Management Journal, Vol 9,pp375-385.

BURRELL, Gibson & MORGAN, Garcth (1979) 'Sociological Paradigms andOrganisational Analysis - Elements of Sociology of Corporate Life', Heinemann,London.

CHRISMAN, James, HOFER, Charles and BOULTON, William (1988) 'Towardsa Systern for Classifying Business Stratcgics', Academy of Management Review,Vol 13, No.3, pp413-428.

COOPER, Harris M. (1984) "The Integrative Research Review', Sage PublicationLtd., London.

DE YOUNG, Garrcu & DAVIS, Dwight (1990) 'Stratcgic Alliances: PiccingTogorher Successful Wortd-Class Partnerships' . Elcctronic Business, Vol 16,NO.lO, pp32-40.

DOWLING, Mclissa (1993) 'Stratcgic Alliances: Is Anybody Really DoingThem?' .Catalog Age, Vol ro, NO.7 pp43-44.

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 No.2 MAY 1995

GROSS, Thomas & NEUMAN, John (1989) 'Strategic Alliances Vital in GlobalMarketing',Marketing News, Vol 23, No.l3, ppl-2.

HAGEDOORN, John & SCHAKENRAAD, Jos (1992) 'Leading Companies andNetworks of Strategic AlIiances in Information Technologies', Research Policy,Vol 21, No.2, ppI63-190.

HAHN, Carl H. (1988) 'Strategic AlIiances: Considerations for the EuropeanAutomotive Industry' ,Executive Speeches, Vol 2, No.6, ppI4-19.

HAMEL, Gary (1991) 'Compctition for Competence and Inter-partner Learningwithin International Strategic AlIiances', Strategic Management Journal, Vol 12,No. July, pp83-103.

RAMEL, Gary, OOZ, Yves, L. and PRAHALAD, C.K. (1989) 'Collaborate withYour Competitors - and Win', Harvard Business Review, Jan/Feb, ppI33-139.

HARRIGAN, Kathryn Rudie (1987) 'Strategic Alliances: Their New Role inGlobal Competition', Colombia Journal of World Business, Vol 22, No.2, pp67-69.

HARRIGAN, Kathryn Rudie (1988a) 'Joint Ventures and Compctitive Strategy',Strategic Management Journal, Vol 9, ppI41-1S8.

HARRIGAN, Kathryn Rudie (1988b) 'Strategic AlIiances and PartnerAsymmetrics' .Management International Review, Vol 28, ppS3-72.

HENNART, Jean-Francois (1988) 'A Transaction Costs Theory of Equity JointVentures' .Strategic Management Journal, Vol 9, pp361-374.

LEI, David & SLOCUM, John W. Jr. (1991) 'Global Strategic AlIianccs: Payoffsand Pitfalls', Organisational Dynamics, Vol 19, No.3, pp44-62.

LEWIS, Jordan D. (1990) 'Making Strategic Alliances Work', Research -Technology Management, No. Nov/Dec, ppI2-1S.

LEWIS, Jordan D. (1992) 'The New Power of Strategic Alliances', PlanningReview, Vol 20, No.5, pp4S-46, 62.

MASaN, Julie Cohen (1993) 'Strategic Alliances: Partriering for Success',Management Review, Vol 82, No.S, pplO-ll+.

MERRIFIELD, Bruce (1992) 'Global Strategic Allianccs Among Firms',International Journal of Technology Management, Vol 7, No.l, pp71-83.

MODIC, Stanley J. (1988) 'Strategic Alliances: A Global Economy DemandsGlobal Partnerships', Industry Week, Vol 237, No.7, pp46-S2.

NARaD, Susan (1988) "Stratcgic Alliances Offer New Options', NationalUnderwriter, Vol 92, No.49, p33.

O'BRIEN, Peter & TULLIS, Mclissa (1989) 'Stratcgic Alliances: The ShirtingBoundaries Between Collaboration and Compctition', Multinational Business,No.4, pplO-17.

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 No.2 MAY 1995

OHMAE, Kenichi (1989) 'The Global Logic of Strategic Alliances', HarvardBusiness Review, March/April, ppI43-154.

OLLEROS, Francisco J. & MACDONALD, Rod (1988) 'Strategic Alliances:Managing Complementarity to Capitalise on Emerging Technologies',Technovation, Vol 7, No.2, ppI55-176.

PARKHE, Arvind (1991) 'Interfirm Diversity, Organisationai Learning, andLongevity in Global Strategic Alliances', Journal of International BusinessStudies, Vol 22, No.4, ppS79-601.

PERLMUTTER, Howard V. and HEENAN, David A. (1986) 'Co-opcrate toCompete Globally', Harvard Business Review, March/April, pp136-152.

PORTER, Michael E. (1987) 'From Competitive Advantage to CorporateStrategy', Harvard Business Review, May/Junc, pp43-S9.

PUCIK, Vladimir (1988) 'Strategic Alliances, Organisationai Learning andCompetitive Advantage: The HRM Agenda', Human Resource Management, Vol27, No.!, pp77-93.

ROBERTS, Roger F. (1992) 'Compctition and Co-operation: The Role of StrategicAlIiances', Executive Speeches, Vol 7, No.2, pp13-1S.

ROSENBAUM, Joe (1993) 'Stratcgic Allianccs in the Global Marketplace",Managing Intellectual Propert y, NoJS, pp17-2S.

SHERMAN, Stratford (1992) 'Are Strategic AlIiances Working?', Fortune. Vol126, No.6, pp77-78.

SIEGEL, Sidney (1956) 'Nonpararnetric Statistics - For the Behavioural Sciences',McGraw-Hill, London.

SPIEGEL, Francis H. Jr. (1993) 'Strategic Alliances, With Care and Creativity',Financial Executive, Vol 9, No.2, pp28-31.

TAKAC, Paul & SINGH, C.P. (1992) 'Strategic Alliances in Banking',Management Decision, Vol 30, No.I, pp32-43.

TURNER, Ian (1988) 'Strategy and Organisation', Management Update -Supplement 10 the Journal o/General Management, Vol 14, No.2, ppl-11.

WElMER, George, KNILL, Bemie, MODIC, Stanley J., & POTTER, Caren (1988)'Irucgratcd Manufacturing VI - Business Without Boundaries ... StrategicAlliances Make Marketing and Manufacturing an International Game', CAE, Vol7, No.! l, ppIM2-IM14.

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 NO.2 MAY 1995

Appendix A: The Analysed Articles

Each article (of 121) is listcd bclow in alphabetical order of Author,with the foIIowing bibliographical data: (year), journal, Vol, No.,pages.For each aniclc the following 7 variables are listcd to the right:1. Auitudc index: -1.00 to 1.00_2. Author profession: acadcmic - industry - consultant - mix (of

academic and non-acadcmic) • indicatcs insufficient information.3. 'Scicntificncss': scicntific - practical,4. Referee procedure: rcferecd - not rcferccd.5. EmpiricaI grounding: Survey - case studies - examples - non-

empirical.6. Theory discussion on SA: yes - no.7. Definition of SA: yes - no.

a s r

-0.58

-0.50

-0.03

-0.58

Anonvmous (1989) Mergers & Acouisitions 24 4 70-71. c n nAbramson (1989) Mauazinc for Mazazine Mrrnr, 18,2,143-147. 0.50 i Il n e n n

OJ3 • D nAnonymous (1990) Dircctors & Boards, 14,3,54-55. i p n n n yAnonymous 1990) Elcctronic Business, 16,6, 5R-60. 0.48 • p n s n n

Anonymous 1993) European Business Rev. 93, 1,41-42. c n nAnonvmous 1993) ENR, 231 17 12. 0.83 • Il n e n n

c n n1.00 • p n

Banks & Baranson (1993) Plannine Rev. 21 6 28-31 +. e V YAnonvmous 1993) Information Today, 10,5,19. 0.67 • Il n

0.94 m p rBaranson 1990) Multinational Business, 2, 54-61. 0.57 • s n e y y

c n nBell (1990) J. of Business Siratccv, 11,6,63-64. c p r n n nBergsman 1993) Nal Real Estatc Investor 35 8 30-40. 0.29 • Il n

c n nBertode (1990) lntl. J. of Tcchnolocv Management, 5,4,375-388. i s r c n nBillon (1986) Business Forum, 11,4,22-25. 0.40 a s n

Brooder & Pritz! (1992) European Management J., \0,4, 412-42t. c V YBorvs & Jcmison (1989) Academy of Mcmt Rev. 14 2 234-249. 0.00 a s r n V y

0.50 a s rBurgers. Hill & Kim (1993) Strategic Management 1., 14,6,419-432. OJ3 a s r s y yCasey (1989) CIO 3 1 57-61. 1.00 • D n e n n

e Y vCellini (1993 Management Accountinz, 74 12,56-59. 0.49 i D n n n n

Cooper & Gardner (1993) In!1. J.. of Phvsical Distr. & Logistics Mzmt, 23 6 14-26. n y y

Glan & Heide (1993) SAM Advanced Management l,58, 4 9-17. OJI a s r0.67 a s r

De Young & Davis (J990) Electronic Business, 16, \O 32-40. e n yCareeIIa (1993) Wall Strcet & Technology, \O \0,30-36. 0.25 • p n n n n

0.50 • p n

Devlin & Bleacklev (1988) Long Rangc Planning, 21, 5,18·23. e y:vDev & Klein 1993) Cornell Hotel & Restaurant Administration Quarterlv 34 1 42-45. 0.03 a s n e n n

0.08 e s rDunlon & Pollard (1986) Chicf Executive 38 38-39 42-45. OJ3 • o n e n nDowline 1993) Catalog Agc JO 7 43-44. 0.46 • p n e n:v

c n nEllion (1987 J. of Information SVSI(~mSManagement. 4 4,57·59. 0.00 i D n n n n

e n nEmond (1992) National Petroleum News, 84, 11,92-96. 0.80 i Il n

Ford (1991) Frontiers of Heahh Service Management, 7, 3, 19-31. n n nRanagan (1993) Management Rev. 82 3 24-26. 0.50 i p n

i s n

Forrest (1990 J. of Small Business Management 28 3 37-45. n V YForrest & Martin (1992) R&D Management, 22, 1,41-53. 0.69 a s r s y y

0.64 a s rGantz (1990 Nctworking Manacement. 8,7,74-78. om c p n e n nGentrv (1993 Intl. l of Purchasina & Materials Mgmt, 29 3 II· 17. 0.89 a s r

Journal of

EUROPEANBUSINESS EDUCATION

Vol.4 NO.2 MAY 1995

Bibliographical dala: Authors (year) journal, Vol, No., pages.

For each article the following 7 variables are listed to the right:1. Auitudc index: -1.00 to 1.00.2. Author profession: acadernic - industry - consultant - mix (of

acadcmic and non-acadcmic), • indicatcs insufficient information.3. 'Scicntificncss': Scicntific - practical.4. Referee procedure: refereed - not refcreed,5. Empirical grounding: Survcy - casc studies - examples - non-

empirical.6. Theory discussion on SA: yes - no.7. Definition of SA: yes - no.

lo(

'-''O

.Sou'O

2.;:

a s r c n n

Halm (1988) Executive Spcechcs, 2, 6 14-19.c n n

0.67 I P n n n n

Hamcl (1991) Straicaic Manazemcnt 1.,12, Julv, 83-103. a s r c y nHaizh (1992) Columhia J. of World Business 27 l 6(}.74. 0.67 a s r

-0.22Harriaarr (1987 Colombia J. of World Business 22, 2, 67-69. 0.77 a s r e n n

e n nHarriaan (1988 Management Int], Rev. 28 special issue 53-72. a s r s y y0.\1Hausrrun (1991) Rcgulaiion, 14 l. 69-76. .{).43 a p nHcllcbust (I9K8) J. of Information Systems Ml!Jllt, 5 1,32-37.

a s r n y yOJ!! c p n e n n

Kaluzny & Zuckerrnan (1993) Hcalthcare Executive, 8, 3, 33-35. n n nKaluznv & Zuckerrnan (1992) Hospital & Hcalth Services Administr. 37 4 477-490. 0.55

0.83 a p nKalu7.0Y, Laccv, Warnecke & Hvres (1993) Hcalth Services Research 28,2,159-182. m s r s y y0.12Kobavashi (198R) Long Ranrc Plannina. 21 2 29-34. a D r1.00 e n n

Kuhn (l9K9) J. of Business Stratcgv, \O 2, 51-53. n n nKooavashi (1990l Tokyo Business Today, 58, 9, 53. 0.83 a s n n n n

0.93 i s rKulkoskv (19R9) Wall Strcct 1. 6 8 18-24 86. 0.69 • p n e n n

• s r e n y

Laveock (1993) Canadian Insurance, 98, 7,24-25. 0.77 I P n e n n

a S r e v vLazarus (1993 Hcalth Care Stratezic Management, 11, II, 14-17.Lei & Slocum 1991) Oraanisational Dvnamics 19 3 44-62.

055 I P n n n n0.54

e n nLci & Slocum 1992) California Manazemcra Rev. 35 l, 81-97. .{).Q9

c D n e n yLeslie 19'13 lne. 15 4,96-104. 0.83 • D nLewis 1990 Rcsearch-Techrotozv Management, 33 6 12-15. 0.79Lewis 1992 Planning Rcv. 20 5,45-46 62 0.43 e p r e n n

a s r e y nLoranzc & Roos 1991) 1. of Busiress Stratcgv, 12 I 25-30. 0.47 a s r e n n

Lowen (1993 Dircct Marketin~, 56, 6, 64-67. n nLoranze Roose & Bronn (1992) Long Ranae Plannina. 25 6 10-17. 0.32

Mackenzic, Ridaway & Sabarese (1986) Chief Executive, 38, 40-4\. n n0.67 I P n c0.50 i p n e

Mazec (1992) Olicf Executive 81 56-61. 0.35 c p n e n n

Mauhcws & Harvev (1988) Ptanning Rev., 16 6,36-41. n nMasen 1993) Management Rev, 82 5, 1G-11t. e n I Y0.53 • D r

McKinncy (1993) Pension World 29 6 11-12. n n0.38 m s r e

I D n1.00 n

n n n

McMillan 1988 Business Qoarterlv, 53 I 35-41.Mcrrifield 1989) Research-Tcchnology Management 32, l, 15-20.

0.67 a s r n n na s n n v n0.8S

Mcrrifield 1992) Irul. 1. of Technology Management 7 1 71-83. 0.780.75

----._. -p ---

I Journal of

EUROPEANBUSINESS EDUCATION

Vo1.4 NO.2 MAY 1995

Bibliographic data: Authors (year) journal, Vol, No., pages.For each article the following 7 variables are listed to the right:1. Attitude index: -1.00 to 1.00.2. Author profession: academic - industry - consultant - mix (of -c

academic and non-academic), • indicates insufficient information. Ul

3. 'Scientificness': scientific - practical. bO I:::I::: o

= Il) :a I:::4. Referee procedure: refereed - not refereed. •....S! ::I I::: .S! -e

5. Empirical grounding: Survey - case studies - examples - non- K V) • V) ] ::I V) enV) II) o V)~ ~ ~ o •... ::I ••...

empirical. 'O 6 8 eo o o.5 o V) =6. Theory discussion on SA: yes - no. lS. r.::: P- ta ;a o~ •.. .~ ~ o

~ .~'O ~ :57. Definition of SA: yes - DO. ::I o I::: '2·E -5 ., ~ P- s r.:::

< ::I 'v ~ E ~<l-e ~ o:: III el,...; N <"i .<f If'i ~ r..:

O'Brien & Tullis (1989) Multinational Business 4 10.17. 0.74 c p n e n nOhmae (989) Harvard Business Rev. 67.2143-154. 0.56 c D r e n nOlive & Mont~omery (1993) Economic Development Review II, I 34-39. 0.46 a p n n n nOlleros & Macdonald (988)Technovation 7 2 155-176. 0.55 a s r n y yPaan (990) Plamina Rev. 18 S 20-22. 0.00 i D r n n yParkhe 1991 1. of Intl. Business Studies, 22, 4,579-601. -0.08 a s r n y yParkhe 1993 Academy of Management 1., 36 4,794-829. 0.00 a s r s Y yParkhe 1993 Organisation Science 4 2 301-324. om a s r s Y yParkhe, Rosenthal & Chandran ()993 Omega, 21, S, 531-39. 0.02 a s r s Y yPaxton (1988) Transoortation & Distribution, 29 i3 32-37. 0.72 i D n c n nPekar ()989) Plannina Rev. 17 4 34-37. 0.25 i s r s n I yPeterson ()993) Marketing News, 27,11,3. 0.62 · D n n n nPhillips ()991) ARMA Records Mgmt Quanerlv, 25, 2,44-48,57. 0.25 · o n n n nPosner (985) Ine. 7 6 74-80. 0.44 · D n c n nPueik (1988) Human Resouree Management, 27 I 77-93. -0.1 1 a s r n y nRandall ()989) Plannina Rev. 17,4 30-33. 0.75 · o r n n nRicks (1993) 1. of Business and IndustriaI Marketing. 8 4 52·57. 0.83 a s n e n 'YRoben (1992 1. of Business Strategy, 13,2,50-53. -0.65 c p r e n nRobens () 992) Executive Speeches, 7 2 13-15. 0.47 t o n e n nRosenbaum ()993) Manaeine Intellectual Propert y, 35 17-25. 0.62 c o n n n nSasaki () 993) Long Ranae Plannina. 26, 6, 41-53. 0.79 i s r c n nSchmerken (1988) Wall Street Computer Rev., 5, 8,24-40,70.71. 0.56 · D n e n nSeymour (1985) Today's Office 20 5 30.36. -0.14 · D n e n nSherman ()992) Fortune, 126,6,77-78. 0.63 · D n e n nSimonson (1990) United States Banker, 100,4,62-63. 0.47 a D n e V VSlowinski (1992) Merzers & Acquisitions 27 1 44-47. 0.17 c s n s v vSoaraco (1993 ) Aviation Week & Space Technology, 138,4,60-61. 0.83 · D n e n nSpiegel 1993 Financial Executive, 9, 2, 28-31. 0.61 i D n c n nTakac & Sinzh (1992) Management Decision 30 I 32-43. 0.37 c s r n Y yTeazarden & Glinow von (1990 Intl. Studies of Mgmt & Organisation, 20 1,99-108. -0.33 a s n n y yTerastra & Simonin (1993) 1. of Int!. Marketing, 1, 1,4-25. 0.12 a s r s Y yThomas & Trevino (1993) 1. of Management Studies 30 5 779-814. 0.19 a s r c n nTurpin 1993) Long Range Planning, 26, 4,1 l-IS. 0.03 a s r e n nWeimer Knill Modic & Potter 1988 CAE,7, II, IM2-IMI4. 0.61 · p n e y yWeimer Knill Modic & Potter 1988 CAE 7 II JMI4-IM20. 0.83 · D n e Y VWeimer, Knill, Modic & Potter 1988 CAE 7 II JM20-JM24. 0.76 · D n e n nWeimer Knill Modic & Potter 1988 CAE 7, 11, IM24-IM30. 0.61 · p n e n nWoods ide & Kandiko (1991) 1. of Euromarketinz. I 1 151-187. -0.17 a s r c n nYablonskv (l990l Plannina Rev., 18,5,18-19. 0.83 i D r c n nYoung 1993 Chemical Week, 152,24, 14. -\.00 · O n e n nZuckerman & Kaluzny (1991) Frontiers of Hcalth Service Mcmt, 7, 3, 3-2\. 0.55 a s n e V y